A Network Pharmacology to Explore the Mechanism of Astragalus Membranaceus in the Treatment of Diabetic Retinopathy

Background Diabetic retinopathy (DR) includes a series of typical lesions affected by retinal microvascular damage caused by diabetes mellitus (DM), which not only seriously damages the vision, affecting the life's quality of patients, but also brings a considerable burden to the family and society. Astragalus Membranaceus (AM) is a commonly used medicine in clinical therapy of eye disorders in traditional Chinese medicine (TCM). In recent years, it is also used for treating DR, but the specific mechanism is unclear. Therefore, this study explores the potential mechanism of AM in DR treatment by using network pharmacology. Methods Based on the oral bioavailability (OB) and drug likeness (DL) of two ADME (absorption, distribution, metabolism, excretion) parameters, Traditional Chinese Medicine Systems Pharmacology Database (TCMSP), Swiss Target Prediction platform, GeneCards, and OMIM database were used to predict and screen the active compounds of AM, the core targets of AM in DR treatment. The Metascape data platform was used to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis on the core targets. Results 24 active compounds were obtained, such as quercetin, kaempferol, and astragaloside IV. There were 169 effective targets of AM in DR treatment, and the targets were further screened and finally, 38 core targets were obtained, such as VEGFA, AKT1, and IL-6. EGFR tyrosine kinase inhibitor resistance, AGE-RAGE signaling pathway in diabetic complications, PI3K-Akt signaling pathway, and other metabolic pathways participated in oxidative stress, cell apoptosis, angiogenesis signal transduction, inflammation, and other biological processes. Conclusion AM treats DR through multiple compounds, multiple targets, and multiple pathways. AM may play a role in the treatment of DR by targeting VEGFA, AKT1, and IL-6 and participating in oxidative stress, angiogenesis, and inflammation.


Introduction
Diabetes mellitus (DM) is a metabolic disorder caused by genetic and environment factors, and it has become a critical health problem worldwide due to its high prevalence and related disability and mortality [1]. In a recent cross-sectional study, it was observed that the prevalence of diabetes had increased slightly among adults living in China from 2007 to 2017 [2]. Paralleling with the increasing pandemic of DM, diabetic retinopathy (DR) is one of the complications of diabetes.
e prevalence of DR is also growing, and an epidemiological study from urban Chinese indicated that the occurrence of DR was 8.1% for patients with DM [3]. DR destroyed the normal function of the retina and was a threat for sight in adults aged 20-74 years [4], which seriously affects the quality of patients' life [5]. Moreover, DR needs repeated treatment, which brings heavy financial burden to the family and society [6]. erefore, early diagnosis and early treatment are particularly important. Clinically, most conventional treatments such as maintaining blood glucose and lipids and keeping a healthy lifestyle are adopted. However, there are certain limitations, so how to effectively prevent and treat DR has become a current research hotspot [7].
Recently, traditional Chinese medicine (TCM) is gradually attracting public's attention owing to its clinical efficacy. Compared with western medicine, TCM is an important alternative strategy because of syndrome differentiation and holistic concept for treating DM [8]. Also, there are various treatment principles for DR, including boosting qi and nourishing yin, fortifying the spleen and removing dampness, invigorating blood, and unblocking the collaterals [9]. rough treatment according to pattern differentiation, herbal medicines have certain advantages in preventing and treating DR [10].
Astragalus Membranaceus (AM) is named "Huangqi" of TCM. AM was originally described in the Sheng Nong's Herbal Classic which is the Classic of Herbal Medicine in ancient TCM practice, which is the dry root of astragalus mongolicus or membranous astragalus [11]. It is one of the popular health-promoting herbs that have been used to strengthen immunity of people for more than 2000 years. Furthermore, it contains numerous constituents, such as saponins, flavonoids, and polysaccharides, which have wide biological activities, for example, anti-inflammatory, antithrombotic, antioxidant, and antidiabetic activities. Chinese herbs have been used since centuries for the treatment and prevention of various disorders. ere were studies that AM was found upregulating insulin-signaling pathways by improving the activity of casein kinase, regulating lipid metabolism, and enhancing insulin resistance to treat T2DM [12]. Owing to properties that they may alleviate several hyperglycaemia-induced pathological occurrences in the retina, AM can potentially be used for the treatment and prevention of DR [13].
ere were numerous studies which reported that AM could downregulate the signaling pathway of p38 MAPK, simultaneously inhibit the mediated pathway of NF-κB, and reduce the production of proinflammatory cytokines, such as IL-1β and TNF-α, which were induced by AGE. Moreover, it can inhibit the formation of retinal angiogenesis by downregulating the level of VEGF in Muller cells induced by hyperglycemia [14].
However, the specific mechanism of action is not clear. In recent years, a new trend emerged by using network pharmacology to understand TCM. From the perspective of systems biology, network pharmacology explores the relationship between drugs-components-targets-diseases, predicts the potential mechanism of drugs to treat diseases, and provides a theoretical basis for effectiveness of TCM to treat the disease [15]. erefore, this study used the method of network pharmacology, collecting data and information from multiple data platforms and exploring the potential mechanism of AM in DR treatment, to provide a theoretical basis for the prevention and treatment of DR by AM. e overall flow chart of the study is shown in Figure 1.

Screening of Potential Compounds and Targets of the
Herb. "Huangqi" (AM) was retrieved in TCMSP and screened with OB ≥ 30% and DL ≥ 0.18. e chemical compounds of AM were obtained, and the compounds of AM from the literature reports were added. All the components of AM were combined. e PubChem platform (https://pubchem.ncbi.nlm.nih.gov/, 2020.6.04) platform was used, and the InCHIKey number or CAS number of the effective components was entered in order to obtain the 2D structural formula of the components. And then, the 2D structural formula was imported into the Swiss Target Prediction platform (http://www.swisstargetprediction.ch/), 2020.6.04 [16]. e species was selected to be "Homo sapiens," and the potential targets of the compounds with the condition of "probability ≥ 0" were predicted. Finally, targets obtained from both databases were combined and deduplicated, and the targets of AM were obtained.
TCMSP: e Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (https://tcmspw.com/tcmsp.php) is a database that is based on the framework of systems pharmacology for herbal medicine, which can predict the compound and target of herbal medicines by setting ADME (absorption, distribution, metabolism, excretion) parameters [17]. ADME parameters: Most herbal medicines are mainly taken orally and undergo metabolic processes, such as metabolism, absorption, distribution, and excretion (ADME) in the human body, including OB and DL.
Oral bioavailability (OB): OB refers to the speed and degree of absorption of the active compounds or active groups of the drug in the systemic circulation, and is a key parameter for evaluating whether the drug can be developed. Generally, molecules with OB ≥ 30% have good oral availability [18].
Drug likeness (DL): the DL of all herbal medicines is acquired by calculation of Tanimoto coefficient. e Tanimoto coefficient is defined as follows [19]: Here, A is the molecular descriptor index of the compounds of herbal medicines to be tested in the TCMSP database and B is the average drug likeness index described by all 6511 small drug molecules in the DrugBank database. e descriptors of all molecules are calculated by the Dragon software (http://www.talete.mi.it/index.htm). Generally, molecules with DL ≥ 0.18 have good drug likeness.
PubChem: PubChem is the world's largest collection platform of chemical information and search information of compounds, such as chemical and physical properties, biological activities, safety, and toxicity information and more and information can be obtained by importing name, molecular formula, structure, and other identifiers of compounds [20].
Swiss Target Prediction: Swiss Target Prediction is a type of online analysis software for predicting targets of small molecules, which can import SMILES and structure of the molecule to predict the targets of small molecule compounds [21].

Screening of Potential Targets of the Disease.
e keyword "diabetic retinopathy" was searched in the Gene-Cards Database (https://www.genecards.org/, 2020.6.05) [22], and the related targets of DR were obtained. In the OMIM database (https://omim.org/,2020.6.05) [23], Gene Map was selected in "Advanced Search" and then the keyword "diabetic retinopathy" was entered, and the related targets of DR were obtained. e targets obtained from the two databases were integrated and duplicated, and the targets of disease were obtained.
Online Mendelian Inheritance in Man (OMIM) and GeneCards are two different databases that provide comprehensive information of all annotated and predicted human genes.

Screening of Core Targets of AM in DR Treatment.
e jvenn online platform (http://jvenn.toulouse.inra.fr/ app/example.html) [24] was used for core targets. e targets of the drug and disease were, respectively, imported, and the intersection targets between drug and disease were obtained. en, the String platform (https://string-db.org/) was used and protein-protein interaction (PPI) was built. In the platform, "Multiple proteins" was selected and the organism was selected as "Homo sapiens." e intersection targets of AM and DR were then imported, the medium confidence of minimum required interaction score was set as 0.4, and the remaining parameters were maintained at the default settings [25]. e file was exported in "TSV format," and then, Cytoscape 3.7.2 was used to analyze the topological properties of the PPI network.
e degree values were calculated, and the core targets of AM in DR treatment with degree ≥ 2 × median were selected. STRING: it is a search tool for the retrieval of interacting genes/proteins, which is applied for predicting the PPI network and detecting the possible relationships.
Cytoscape: it is a type of open-source software used to analyze and visualize biological networks [26].

GO and KEGG Pathway Enrichment Analysis.
e core targets of AM in DR treatment were imported into the Metascape platform, and in this, species was selected as "H sapiens," "Custom analysis" was selected, and parameters of P value < 0.01, a minimum count of 3, and an enrichment factor >1.5 (the enrichment factor is the ratio between the observed counts and the counts expected by chance) were set. GO and KEGG pathway enrichment analyses on core targets were performed.
Metascape: it is a gene function annotation analysis tool, to realize the recognition of gene or protein function through the bioinformatics analysis of genes and proteins [27].
Gene Ontology (GO): it is a bioinformatics tool used for annotating genes, gene products, and sequences to analyze their biological processes. GO includes 3 aspects: molecular function (MF); biological process (BP); cellular component (CC).
KEGG: it is a database resource comprising information on the biological systems of genome sequences and other high-throughput data [28]. Pathway-target network Step 3 Network construction GO and KEGG analysis Step 2 Target analysis Known potential targets of DR Genecards and OMIM Step 1: Target prediction Diabetic retinopathy (DR)

Active Compounds and Targets of the Drug.
On the TCMSP platform, the ADME parameters of OB ≥ 30% and DL ≥ 0.18 were selected 20 active compounds were obtained, and then, according to the relevant literature [29], additional 4 active compounds were obtained and merged with the former, and finally, a total of 24 compounds of AM were identified (Table 1).
2D structural formulas of 24 compounds were obtained from PubChem, and then, compounds were imported into the Swiss Target Prediction platform to obtain the target information of 24 compounds. Targets were merged and deduplicated, and a total of 482 targets of AM were obtained. e component-target network (C-T network) was constructed ( Figure 2) by using Cytoscape 3.7.2, and there were 506 nodes and 1310 edges. e function of "Network Analyzer" in Cytoscape 3.7.2 was used to perform topological analysis on the components (Table 2), and the main compounds were jaranol, isoflavanone, kaempferol, and quercetin.

Core Targets of AM on DR.
ere were 217 targets of DR in the OMIM database, while there were 1753 targets of DR in the GeneCards database. e targets from the two databases were merged, the duplicates were removed, and finally, 1901 targets of DR were obtained. 482 targets of AM and 1901 targets of DR were imported into Venny2.1.0 online platform, and there were 169 common targets ( Figure 3). 169 targets were imported into STRING online platform, and a file in "TSV" format was exported. en, it was imported into Cytoscape 3.7.2, the function of "network analyzer" was used, and the median of degree was set as 18. 38 core targets of AM in DR treatment with degree ≥ 36 (2 × 18) were screened (Figure 4).

PPI of Core
Targets. 38 core targets were imported into the STRING platform, the PPI of the core targets were obtained ( Figure 5), and finally, there were 38 nodes and 555 edges. e average node degree was 29.2. e specific information of targets is shown in Table 3.

Enrichment Analysis of 38 Targets.
ere were 38 targets of AM on DR; the biological function of the 38 targets was elucidated. GO analysis and KEGG pathway analysis on 38 targets were performed. Based on GO results, there were 79 processes of molecular functions, such as protein kinase activity, protein serine/threonine kinase activity, and kinase binding; there were 418 biological processes, such as transmembrane receptor protein tyrosine kinase signaling pathway, positive regulation of cell migration, and positive regulation of cell motility; there were 42 cellular components, such as membrane raft, membrane microdomain, and membrane region. In addition, there were 123 pathways for results of KEGG. e top 20 results of KEGG were listed and the scatter plot was drawn. e larger the dot, the more the count, the redder the color, and the smaller the P value (Figure 6), and EGFR tyrosine kinase inhibitor resistance, AGE-RAGE signaling pathway in diabetic complications, PI3K-AKT signaling pathway and IL-17 signaling pathway may be involved in the treatment of DR by AM. e top 20 pathways were screened and the target-pathway network (T-P network) was constructed (Figure 7). ere were 55 nodes and 331 edges, and the topology analysis was further analyzed. e main pathways included EGFR tyrosine kinase inhibitor resistance, AGE-RAGE signaling pathway in diabetic complications, and PI3K-Akt signaling pathway (Table 4).

Discussion
DR is a common and severe complication of DM, which is the main cause of blindness in adults. In the early stage, the high-glucose environment influences the retinal microvascular damage, which is characterized by loss of retinal cells, thickening of the basement membrane of capillaries, expansion and deformation of capillaries, and destruction of the blood-retinal barrier. Moreover, the progression is characterized by the occlusion of capillaries and the formation of pathological neovascularization, retinal hemorrhage, etc, which cause irreversible loss of vision [30]. Experts and scholars have been devoted to the research of effective prevention and treatment of DR, but there are still many unsolved mysteries. In recent years, studies reported that the treatment of this disease by TCM had been increasing year by year and the treatment efficiency had gradually improved. AM is a class of herbs used for boosting qi and invigorating blood, but its precise mechanism of AM on DR is not yet clear. erefore, it is necessary to explore the mechanism of AM in DR treatment by using functions of screening compounds, predicting targets, and analyzing pathways of drugs in network pharmacology.
In this study, we firstly identified the active compounds of AM and combined literature reviews. 24 compounds were identified, and the specific information of compounds is shown in Table 1. Some compounds, such as quercetin, kaempferol, and astragaloside IV were reported to have played an important role in DR. Quercetin could reduce the expression of MMP-9 and VEGF in rats with diabetic retinopathy [31] and inhibit the generation of angiogenesis both in vitro and ex vivo [32]. Quercetin can reduce the overproduction of MCP, IL-6, and ROS and downregulate the expression of P53, 4 Evidence-Based Complementary and Alternative Medicine Bax, and Caspase-3 by promoting PTEN/AKT pathway and inhibiting NF-κB pathway, which could protect ARPE-19 cell damage induced by high glucose [33]. Kaempferol could protect against oxidative stress leading to damage of human RPE cells through its antioxidant activity and antiapoptosis function [34]. Moreover, it could suppress angiogenesis of human retinal endothelial cells (HRECs) via targeting VEGF and PGF to inhibit the activation of Src-Akt1-Erk1/2 signaling pathway under glucose conditions [35]. Astragaloside IV could prevent  Evidence-Based Complementary and Alternative Medicine 5 the activation of ERK1/2 phosphorylation and NF-κB and further relieve the RGCs dysfunction in ab/db mice with DR [36], and it has potential protective effects on retinal capillary endothelial cells (RCECs) incubated with high glucose conditions [37]. We predicated 169 common targets between AM and DR, and on further screening, 38 core targets for AM in DR treatment were obtained. e targets were mainly enriched for oxidative stress, cell proliferation, apoptosis, inflammation, etc. When it comes to core targets, Akt1, VEGFA, IL-6, Caspase-3, and STAT3 played an important role in the process of AM in DR treatment. e detailed information is given in Table 2. VEGFA could be an activator of the stimulation of angiogenesis, and particularly, it was reported to be strongly upregulated in PDR [38]. Numerous studies had reported that AKT signaling pathway could be involved in the occurrence and development of DR and the activation of AKT can maintain the steady state of endothelial function of retinal capillaries [39]. Caspase-3 could participate in the pathological process of DR by regulating apoptosis, and the abnormally high expression level of Caspase-3 may be correlated with the severity of DR [40]. e inhibition of STAT13 can alleviate retinal inflammation and reduce retinal cell death in rats of DR, while IL-6 could also be a proinflammatory factor, and and by inhibiting the activation of IL-6/STAT3 signaling pathway and then affecting the oxidation and antioxidation process of the retina, the purpose of protecting the retina of diabetic rats can be achieved [41].    : PPI of core targets. e layout of the outer ring went in an anticlockwise direction according to the area and color of the nodes, andthe yellow nodes represent the core targets of AM on DR; the size and transparency of the nodes were divided according to the degree value from large to small. e inner edges represent the relationship between the core targets. e more the edges of the nodes were, the larger the area of the nodes, the darker the color, and the more important the targets.
Subsequently, we compared the results of GO and KEGG enrichment analyses for 38 key targets, and the results indicated that regulation of various kinase activities, response to insulin, and regulation of vascular endothelial cells played a main role in the process of AM in DR treatment. Furthermore, in delaying the progression of diabetic retinopathy, oxidative stress and inflammation could also play an important role [42]. e action pathway mainly includes EGFR tyrosine kinase inhibitor resistance, AGE-RAGE signaling pathway in diabetic complications, and PI3K-Akt signaling pathways. Previous studies had indicated that renal epidermal growth factor receptors (EGFRs) were activated in models of diabetic nephropathy (DN), and inhibition of EGFR activity protected against progressive DN in T1 DM and T2 DM [43]. So the EGFR tyrosine kinase inhibitor resistance may delay the progression of diabetes and improve complications such as diabetic retinopathy. In the condition of high glucose, nonenzymatic glycosylation of proteins and lipids can be induced, advanced glycation endproducts (AGEs) can be generated, and through the activation of AGE receptor (RAGE), oxidative stress and inflammation can be induced [44], which could damage the retina. e details of AM on the AGEs-RAGE pathway are shown in Figure 8, and it was cited from the KEGG database (https://www.kegg.jp/).
In addition, there were several pathways, such as pathway in cancer, proteoglycans in cancer, and hepatitis bare, indicating AM might exhibit potential therapeutic effects on these diseases, while demonstrating that the progression of DR is influenced by other diseases. ere were multiple targets, multiple pathways, and multiple processes involved in treatment of DR through AM. In addition, there were still some deficiencies in this study. e network pharmacology method was used to predict the   : T-P network. e layout of the outer and inner ring went in an anticlockwise direction according to the area and color of the nodes; the green nodes represent the pathways while the blue nodes represent targets. e size and transparency of the nodes were divided according to the degree value from large to small. e inner edges represented the relationship between the pathways and the targets. e more edges, the larger the area of the nodes, the darker of the color, and the more important the nodes. 8 Evidence-Based Complementary and Alternative Medicine targets and pathways of AM in DR treatment, which still need to be verified by future pharmacological studies.

Conclusions
In summary, we found that AM in DR treatment has the characteristics of multiple components, multiple targets, and multiple pathways through network pharmacology. AGE-RAGE signaling pathway in diabetic complications and PI3K-Akt signaling pathway might play an important role in the treatment process, providing a theoretical basis for AM in DR treatment and a new direction for the development of drugs for DR.
Data Availability e data of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.

Authors' Contributions
QJ, XFH, and LKX conceived and designed the study; QJ and JX wrote the paper; MS and HY organized the data, SHW, GPW, and MLM analyzed the data. All authors had read and approved the final manuscript.  Evidence-Based Complementary and Alternative Medicine 9